Our work focuses on several different synapses and cell types in the inner retina. In retinal ganglion cells (RGCs), which integrate synaptic input from the network and relay the visual signal to the rest of the brain, we find that certain types of NMDA-type glutamate receptors (NMDARs) are localized specifically to limit their activation under certain conditions. Anatomical work in the lab shows that NMDARs are localized primarily perisynaptically at ON synapses, while they are localized in the postsynaptic density at OFF synapses (Zhang and Diamond, 2006). Moreover, NMDARs containing the NR2B subunit are more likely to be perisynaptically targeted and at ON synapses, while NR2A-containing NMDARs are primarily synaptic and at OFF synapses. This contrasting subsynaptic distribution correlates with the expression pattern of scaffolding proteins known to be important for glutamate receptor targeting (manuscript in preparation). We find that PSD-95 is co-localized with NR2A-containing receptors, while the expression of SAP102 is more aligned with NR2B containing receptors. Interestingly, we also find that specific splice variants of the NR1 subunit co-localize with specific NR2 subunits. Our physiological results support these anatomical findings, as synaptically-evoked ON responses are mediated by a greater fraction of NR2B-containing receptors than OFF synapses (manuscript in preparation). We also find that the fractional contribution of different subunits can be modulated by synaptic release of NMDAR coagonist, either glycine or serine. These results indicate that NMDARs may play distinct roles in synaptic processing of light increments and decrements in the visual world. We are examining this idea in greater detail by measuring synaptic Ca signals mediated by NMDARs in ganglion cell dendrites. Our initial results indicate that synaptic, NMDAR-mediated Ca transients exhibit different dynamics than those due to influx through voltage-gated calcium channels. We are in the process of testing the relative contribution of intracellular calcium stores and extrusion mechanisms in the dynamics of dendritic calcium. We also have expanded our study of inhibitory synaptic connections made by amacrine cells within the inner retina, to understand how feedforward and feedback inhibition contributes to signal processing in this network. We find that A17 amacrine cells provide rapid GABAergic feedback to rod bipolar cell terminals via a release process that is independent of membrane depolarization or voltage-gated calcium channels (Chavez, et al., 2006). This rapid feedback, driven by activation of calcium-permeable AMPA receptors in the A17 amacrine cell, may be essential to prevent the rapid depletion of readily-releasable vesicles from the rod bipolar cell synaptic terminal (Singer and Diamond, 2006). More recent work (two papers nearing submission) indicate that two other types of feedback inhibition onto bipolar cells exhibit different characteristics and mechanisms of modulation. In addition to providing valuable information about feedback, this work is enabling us to begin to make functional sense of the vast array of amacrine cells (>two dozen cell types) in the mammalian retina. Finally, in another project we are combining electrophysiological and imaging techniques to explore the mechanisms of signal integration in A17 amacrine cells. We find that active, calcium-dependent conductances may be essential to compartmentalize signal processing within A17 dendrites, forming a crucial component of the rod pathway that underlies scotopic (low light) vision in the mammalian retina.